Skip to content

Commit

Permalink
outsourced thelis DHT flush class into own file
Browse files Browse the repository at this point in the history 
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@1706 6c8d7289-2bf4-0310-a012-ef5d649a1542
  • Loading branch information
orbiter committed Feb 19, 2006
1 parent aa4b04e commit 73dad68
Show file tree
Hide file tree
Showing 3 changed files with 42 additions and 276 deletions.
29 changes: 13 additions & 16 deletions htroot/IndexTransfer_p.java
View file
Expand Up @@ -55,7 +55,6 @@

import de.anomic.http.httpHeader;
import de.anomic.plasma.plasmaSwitchboard;
import de.anomic.plasma.plasmaWordIndexDistribution;
import de.anomic.server.serverObjects;
import de.anomic.server.serverSwitch;
import de.anomic.yacy.yacyCore;
Expand All @@ -75,32 +74,30 @@ public static serverObjects respond(httpHeader header, serverObjects post, serve
prop.put("running_status","Disconnected peer");
} else {
boolean deleteIndex = post.get("deleteIndex", "0").equals("1");
switchboard.indexDistribution.startTransferWholeIndex(seed,deleteIndex);
switchboard.startTransferWholeIndex(seed,deleteIndex);
prop.put("LOCATION","");
return prop;
}
} else if (post.containsKey("stopIndexTransfer")) {
switchboard.indexDistribution.stopTransferWholeIndex(true);
switchboard.stopTransferWholeIndex(true);
prop.put("LOCATION","");
return prop;

} else if (post.containsKey("newIndexTransfer")) {
switchboard.indexDistribution.abortTransferWholeIndex(true);
switchboard.abortTransferWholeIndex(true);
prop.put("LOCATION","");
return prop;
}
}


// insert constants
plasmaWordIndexDistribution.transferIndexThread transfThread = switchboard.indexDistribution.transferIdxThread;
prop.put("wcount", Integer.toString(switchboard.wordIndex.size()));
prop.put("ucount", Integer.toString(switchboard.urlPool.loadedURL.size()));
prop.put("running",(transfThread==null)?0:1);
if (transfThread != null) {
String[] status = transfThread.getStatus();
String[] range = transfThread.getRange();
int[] chunk = transfThread.getIndexCount();
prop.put("running",(switchboard.transferIdxThread==null)?0:1);
if (switchboard.transferIdxThread != null) {
String[] status = switchboard.transferIdxThread.getStatus();
String[] range = switchboard.transferIdxThread.getRange();
int[] chunk = switchboard.transferIdxThread.getIndexCount();

prop.put("running_selection.status",status[0]);
prop.put("running_selection.twrange", range[0]);
Expand All @@ -112,13 +109,13 @@ public static serverObjects respond(httpHeader header, serverObjects post, serve


//prop.put("running_twEntityCount",transfThread.getTransferedEntityCount());
prop.put("running_twEntryCount",transfThread.getTransferedEntryCount());
prop.put("running_twEntryCount",switchboard.transferIdxThread.getTransferedEntryCount());
//prop.put("running_twEntityPercent",Float.toString(transfThread.getTransferedEntityPercent()));
prop.put("running_twEntitySpeed",Integer.toString(transfThread.getTransferedEntitySpeed()));
prop.put("running_twEntitySpeed",Integer.toString(switchboard.transferIdxThread.getTransferedEntitySpeed()));

prop.put("running_deleteIndex", transfThread.deleteIndex()?1:0);
prop.put("running_peerName",transfThread.getSeed().getName());
prop.put("running_stopped",(transfThread.isFinished()) || (!transfThread.isAlive())?1:0);
prop.put("running_deleteIndex", switchboard.transferIdxThread.deleteIndex()?1:0);
prop.put("running_peerName",switchboard.transferIdxThread.getSeed().getName());
prop.put("running_stopped",(switchboard.transferIdxThread.isFinished()) || (!switchboard.transferIdxThread.isAlive())?1:0);
} else {
if (!prop.containsKey("running_status")) prop.put("running_status","Not running");
}
Expand Down
28 changes: 28 additions & 0 deletions source/de/anomic/plasma/plasmaSwitchboard.java
View file
Expand Up @@ -201,6 +201,7 @@ public final class plasmaSwitchboard extends serverAbstractSwitch implements ser
//public StringBuffer crl; // local citation references
public StringBuffer crg; // global citation references
public dbImportManager dbImportManager;
public plasmaDHTFlush transferIdxThread = null;

/*
* Remote Proxy configuration
Expand Down Expand Up @@ -781,6 +782,7 @@ public int htEntrySize() {
public void close() {
log.logConfig("SWITCHBOARD SHUTDOWN STEP 1: sending termination signal to managed threads:");
terminateAllThreads(true);
if (transferIdxThread != null) stopTransferWholeIndex(false);
log.logConfig("SWITCHBOARD SHUTDOWN STEP 2: sending termination signal to threaded indexing");
// closing all still running db importer jobs
this.dbImportManager.close();
Expand Down Expand Up @@ -1905,6 +1907,32 @@ public boolean verifyAuthentication(httpHeader header, boolean strict) {
return false;
}

public void startTransferWholeIndex(yacySeed seed, boolean delete) {
if (transferIdxThread == null) {
this.transferIdxThread = new plasmaDHTFlush(this.log, this.wordIndex, seed, delete);
this.transferIdxThread.start();
}
}

public void stopTransferWholeIndex(boolean wait) {
if ((transferIdxThread != null) && (transferIdxThread.isAlive()) && (!transferIdxThread.isFinished())) {
try {
this.transferIdxThread.stopIt(wait);
} catch (InterruptedException e) { }
}
}

public void abortTransferWholeIndex(boolean wait) {
if (transferIdxThread != null) {
if (!transferIdxThread.isFinished())
try {
this.transferIdxThread.stopIt(wait);
} catch (InterruptedException e) { }
transferIdxThread = null;
}
}


public void terminate() {
this.terminate = true;
this.shutdownSync.V();
Expand Down
261 changes: 1 addition & 260 deletions source/de/anomic/plasma/plasmaWordIndexDistribution.java
View file
Expand Up @@ -68,7 +68,6 @@ public final class plasmaWordIndexDistribution {
private boolean closed;
private boolean gzipBody4Distribution;
private int timeout4Distribution;
public transferIndexThread transferIdxThread = null;

public plasmaWordIndexDistribution(
plasmaURLPool urlPool,
Expand Down Expand Up @@ -108,9 +107,7 @@ public void disableWhileCrawling() {

public void close() {
closed = true;
if (transferIdxThread != null) {
stopTransferWholeIndex(false);
}

}

private boolean isClosed() {
Expand Down Expand Up @@ -278,260 +275,4 @@ void closeTransferIndexes(plasmaWordIndexEntryContainer[] indexContainers) {
}
}

public void startTransferWholeIndex(yacySeed seed, boolean delete) {
if (transferIdxThread == null) {
this.transferIdxThread = new transferIndexThread(seed,delete);
this.transferIdxThread.start();
}
}

public void stopTransferWholeIndex(boolean wait) {
if ((transferIdxThread != null) && (transferIdxThread.isAlive()) && (!transferIdxThread.isFinished())) {
try {
this.transferIdxThread.stopIt(wait);
} catch (InterruptedException e) { }
}
}

public void abortTransferWholeIndex(boolean wait) {
if (transferIdxThread != null) {
if (!transferIdxThread.isFinished())
try {
this.transferIdxThread.stopIt(wait);
} catch (InterruptedException e) { }
transferIdxThread = null;
}
}

public class transferIndexThread extends Thread {
private yacySeed seed = null;
private boolean delete = false;
private boolean finished = false;
private boolean gzipBody4Transfer = false;
private int timeout4Transfer = 60000;
private int transferedEntryCount = 0;
private int transferedContainerCount = 0;
private String status = "Running";
private String oldStartingPointHash = "------------", startPointHash = "------------";
private int initialWordsDBSize = 0;
private int chunkSize = 500;
private final long startingTime = System.currentTimeMillis();
private final plasmaSwitchboard sb;
private plasmaDHTTransfer worker = null;

public transferIndexThread(yacySeed seed, boolean delete) {
super(new ThreadGroup("TransferIndexThreadGroup"),"TransferIndex_" + seed.getName());
this.seed = seed;
this.delete = delete;
this.sb = plasmaSwitchboard.getSwitchboard();
this.initialWordsDBSize = sb.wordIndex.size();
this.gzipBody4Transfer = "true".equalsIgnoreCase(sb.getConfig("indexTransfer.gzipBody","false"));
this.timeout4Transfer = (int) sb.getConfigLong("indexTransfer.timeout",60000);
//this.maxOpenFiles4Transfer = (int) sb.getConfigLong("indexTransfer.maxOpenFiles",800);
}

public void run() {
performTransferWholeIndex();
}

public void stopIt(boolean wait) throws InterruptedException {
this.finished = true;
if (wait) this.join();
}

public boolean isFinished() {
return this.finished;
}

public boolean deleteIndex() {
return this.delete;
}

public int[] getIndexCount() {
plasmaDHTTransfer workerThread = this.worker;
if (workerThread != null) {
return new int[]{this.chunkSize, workerThread.getIndexCount()};
}
return new int[]{this.chunkSize, 500};
}

public int getTransferedEntryCount() {
return this.transferedEntryCount;
}

public int getTransferedContainerCount() {
return this.transferedContainerCount;
}

public float getTransferedContainerPercent() {
long currentWordsDBSize = sb.wordIndex.size();
if (initialWordsDBSize == 0) return 100;
else if (currentWordsDBSize >= initialWordsDBSize) return 0;
//else return (float) ((initialWordsDBSize-currentWordsDBSize)/(initialWordsDBSize/100));
else return (float)(this.transferedContainerCount*100/initialWordsDBSize);
}

public int getTransferedEntitySpeed() {
long transferTime = System.currentTimeMillis() - startingTime;
if (transferTime <= 0) transferTime = 1;
return (int) ((1000 * transferedEntryCount) / transferTime);
}

public yacySeed getSeed() {
return this.seed;
}

public String[] getStatus() {
plasmaDHTTransfer workerThread = this.worker;
if (workerThread != null) {
return new String[]{this.status,workerThread.getStatusMessage()};
}
return new String[]{this.status,"Not running"};
}

public String[] getRange() {
plasmaDHTTransfer workerThread = this.worker;
if (workerThread != null) {
return new String[]{"[" + oldStartingPointHash + ".." + startPointHash + "]",workerThread.getRange()};
}
return new String[]{"[" + oldStartingPointHash + ".." + startPointHash + "]","[------------..------------]"};
}

public void performTransferWholeIndex() {
plasmaDHTChunk newDHTChunk = null, oldDHTChunk = null;
try {
// pausing the regular index distribution
// TODO: adding sync, to wait for a still running index distribution to finish
plasmaWordIndexDistribution.this.paused = true;

// initial startingpoint of intex transfer is "------------"
plasmaWordIndexDistribution.this.log.logFine("Selected hash " + startPointHash + " as start point for index distribution of whole index");

/* Loop until we have
* - finished transfer of whole index
* - detected a server shutdown or user interruption
* - detected a failure
*/
long selectionStart = System.currentTimeMillis(), selectionEnd = 0, selectionTime = 0, iteration = 0;

while (!finished && !Thread.currentThread().isInterrupted()) {
iteration++;
selectionStart = System.currentTimeMillis();
oldDHTChunk = newDHTChunk;

// selecting 500 words to transfer
this.status = "Running: Selecting chunk " + iteration;
newDHTChunk = new plasmaDHTChunk(plasmaWordIndexDistribution.this.log, wordIndex, sb.urlPool.loadedURL, this.chunkSize/3, this.chunkSize, this.startPointHash);

/* If we havn't selected a word chunk this could be because of
* a) no words are left in the index
* b) max open file limit was exceeded
*/
if ((newDHTChunk == null) ||
(newDHTChunk.containerSize() == 0) ||
(newDHTChunk.getStatus() == plasmaDHTChunk.chunkStatus_FAILED)) {
if (sb.wordIndex.size() > 0) {
// if there are still words in the index we try it again now
startPointHash = "------------";
} else {
// otherwise we could end transfer now
plasmaWordIndexDistribution.this.log.logFine("No index available for index transfer, hash start-point " + startPointHash);
this.status = "Finished. " + iteration + " chunks transfered.";
finished = true;
}
} else {

// getting start point for next DHT-selection
oldStartingPointHash = startPointHash;
startPointHash = newDHTChunk.lastContainer().wordHash(); // DHT targets must have greater hashes

selectionEnd = System.currentTimeMillis();
selectionTime = selectionEnd - selectionStart;
plasmaWordIndexDistribution.this.log.logInfo("Index selection of " + newDHTChunk.indexCount() + " words [" + newDHTChunk.firstContainer().wordHash() + " .. " + newDHTChunk.lastContainer().wordHash() + "]" +
" in " +
(selectionTime / 1000) + " seconds (" +
(1000 * newDHTChunk.indexCount() / (selectionTime+1)) + " words/s)");
}

// query status of old worker thread
if (worker != null) {
this.status = "Finished: Selecting chunk " + iteration;
worker.join();
if (!worker.success) {
// if the transfer failed we abort index transfer now
this.status = "Aborted because of Transfer error:\n" + worker.getStatus();

// abort index transfer
return;
} else {
/*
* If index transfer was done successfully we close all remaining open
* files that belong to the old index chunk and handover a new chunk
* to the transfer thread.
* Addintionally we recalculate the chunk size to optimize performance
*/

this.chunkSize = worker.getIndexCount();
long transferTime = worker.getTransferTime();
//TODO: only increase chunk Size if there is free memory left on the server

// we need aprox. 73Byte per IndexEntity and an average URL length of 32 char
//if (ft.freeMemory() < 73*2*100)
if (transferTime > 60*1000) {
if (chunkSize>200) chunkSize-=100;
} else if (selectionTime < transferTime){
this.chunkSize +=100;
//chunkSize+=50;
} else if (selectionTime >= selectionTime){
if (chunkSize>200) chunkSize-=100;
}

selectionStart = System.currentTimeMillis();

// deleting transfered words from index
if (delete) {
this.status = "Running: Deleting chunk " + iteration;
transferedEntryCount += oldDHTChunk.indexCount();
transferedContainerCount += oldDHTChunk.containerSize();
int urlReferences = oldDHTChunk.deleteTransferIndexes();
plasmaWordIndexDistribution.this.log.logFine("Deleted from " + oldDHTChunk.containerSize() + " transferred RWIs locally " + urlReferences + " URL references");
} else {
transferedEntryCount += oldDHTChunk.indexCount();
transferedContainerCount += oldDHTChunk.containerSize();
}
oldDHTChunk = null;
}
this.worker = null;
}

// handover chunk to transfer worker
if ((newDHTChunk != null) &&
(newDHTChunk.containerSize() > 0) ||
(newDHTChunk.getStatus() == plasmaDHTChunk.chunkStatus_FILLED)) {
worker = new plasmaDHTTransfer(log, seed, newDHTChunk,
gzipBody4Transfer, timeout4Transfer, iteration,
startPointHash, oldStartingPointHash);
worker.start();
}
}

// if we reach this point we were aborted by the user or by server shutdown
if (sb.wordIndex.size() > 0) this.status = "aborted";
} catch (Exception e) {
this.status = "Error: " + e.getMessage();
plasmaWordIndexDistribution.this.log.logWarning("Index transfer to peer " + seed.getName() + ":" + seed.hash + " failed:'" + e.getMessage() + "'",e);

} finally {
if (worker != null) {
worker.stopIt();
try {worker.join();}catch(Exception e){}
// worker = null;
}

plasmaWordIndexDistribution.this.paused = false;
}
}

}

}

0 comments on commit 73dad68

Please sign in to comment.